2306 Commits

Author SHA1 Message Date
aoiasd
dcf04a58b8
feat: support use score function on segment search and use filter (#43868)
relate: https://github.com/milvus-io/milvus/issues/43867
Support boost function score, multiply by the weight if match filter.

Signed-off-by: aoiasd <zhicheng.yue@zilliz.com>
2025-08-19 16:15:45 +08:00
Gao
b602b4187d
enhance: upgrade aws-sdk from 1.9.234 to 1.11.352 (#43916)
issue: #43908

Signed-off-by: chasingegg <chao.gao@zilliz.com>
2025-08-19 11:11:45 +08:00
yihao.dai
64ab3d2681
enhance: Improve error message when query vector and dim mismatch (#43835)
/kind improvement

Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
2025-08-18 01:07:44 +08:00
foxspy
647c2bca2d
enhance: Support streaming read and write of vector index files (#43824)
issue: #42032

Signed-off-by: xianliang.li <xianliang.li@zilliz.com>
2025-08-15 23:41:43 +08:00
Alexander Guzhva
ebb10dfae0
enhance: better auto-detect of SVE options for the bitset library (#43833)
Enables the compilation of SVE code for the bitset library if a C++
compiler supports it.

There are two conditions for enabling the SVE code
* a C++ compiler needs to have a `-march=armv8-a+sve`
* `arm_sve.h` header must be available

AFAIK, `gcc 7 does not support SVE`, `gcc 8` and `gcc 9` support SVE,
but have no `arm_sve.h` file, and only `gcc 10` has both.

Signed-off-by: Alexandr Guzhva <alexanderguzhva@gmail.com>
2025-08-15 22:37:44 +08:00
sthuang
5e4eb4a6e0
enhance: [StorageV2] bump storage version (#43871)
related: https://github.com/milvus-io/milvus/issues/43869

bump storage version. include the following feature:
* https://github.com/milvus-io/milvus-storage/pull/231
* https://github.com/milvus-io/milvus-storage/pull/232
* https://github.com/milvus-io/milvus-storage/pull/233

Signed-off-by: shaoting-huang <shaoting.huang@zilliz.com>
2025-08-15 17:59:43 +08:00
sthuang
c102fa8b0b
enhance: [StorageV2] zero copy for packed writer record batch (#43779)
The Out of Memory (OOM) error occurs because a handler retains the
entire ImportRecordBatch in memory. Consequently, even when child arrays
within the batch are flushed, the memory for the complete batch is not
released. We temporarily fixed by deep copying record batch in #43724.

The proposed fix is to split the RecordBatch into smaller sub-batches by
column group. These sub-batches will be transferred via CGO, then
reassembled before being written to storage using the Storage V2 API.
Thus we can achieve zero-copy and only transferring references in CGO.

related: #43310

Signed-off-by: shaoting-huang <shaoting.huang@zilliz.com>
2025-08-15 10:11:44 +08:00
congqixia
f032044125
enhance: Refine segcore param change callback (#43838)
Related to #43230

This PR
- Move segcore setup function to `initcore` package to remove cgo
dependency from pkg
- Register core callback only for components depends on segcore
- Rectify `UpdateLogLevel` implementation

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-08-13 19:31:44 +08:00
zhagnlu
b7c7df9440
fix: fix delete consumer bug for cocurrency R-W (#43831)
#41570

Signed-off-by: luzhang <luzhang@zilliz.com>
Co-authored-by: luzhang <luzhang@zilliz.com>
2025-08-12 11:37:42 +08:00
Gao
81a0915c29
enhance: add milvus-common module to decouple knwhere & segcore (#43624)
issue: https://github.com/milvus-io/milvus/issues/42032
https://github.com/milvus-io/milvus/issues/41435

based on pr: https://github.com/milvus-io/milvus/pull/42124

---------

Signed-off-by: chasingegg <chao.gao@zilliz.com>
Co-authored-by: xianliang.li <xianliang.li@zilliz.com>
2025-08-11 14:09:42 +08:00
zhagnlu
5b83975d39
enhance:convert multi not equal to not in (#43690)
#43689

Signed-off-by: luzhang <luzhang@zilliz.com>
Co-authored-by: luzhang <luzhang@zilliz.com>
2025-08-08 10:37:40 +08:00
sparknack
169be30a76
enhance: cachinglayer: reserve resource for inevictable cachecell (#43602)
issue: #41435

---------

Signed-off-by: Shawn Wang <shawn.wang@zilliz.com>
2025-08-08 10:35:49 +08:00
zhagnlu
c04d678ad4
enhance: make segcore params effective without restarting milvus (#43231)
#43230

Signed-off-by: luzhang <luzhang@zilliz.com>
Co-authored-by: luzhang <luzhang@zilliz.com>
2025-08-08 10:33:48 +08:00
congqixia
1561a4ae8c
enhance: [StorageV2] Avoid create local parent dir if fs remote (#43790)
Related to #43752
milvus-storage pr: milvus-io/milvus-storage#230

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-08-08 10:19:40 +08:00
congqixia
b6199acb05
enhance: Utilize search_batch_pks for search_ids of PkTerm (#43751)
Related to #43660

---------

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-08-07 14:19:40 +08:00
congqixia
d414f6bd4d
enhance: Add assertion preventing reload same field (#43736)
Related to #43725

This patch add assertion preventing segment reloading same field column.
Also improve the message info when pk already exists.

---------

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-08-05 19:35:39 +08:00
yihao.dai
cb7be8885d
enhance: Deep copy arraw array (#43724)
Deep copy arrow array and make a new RecordBatch with the copied array.

issue: https://github.com/milvus-io/milvus/issues/43310

---------

Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
2025-08-05 00:31:38 +08:00
Chun Han
d826d6ac91
fix: try to get span raw data for variable length data type(#43544) (#43705)
related: #43544

Signed-off-by: MrPresent-Han <chun.han@gmail.com>
Co-authored-by: MrPresent-Han <chun.han@gmail.com>
2025-08-04 11:15:38 +08:00
aoiasd
4f02b06abc
enhance: support set lindera dict build dir and download url in yaml (#43541)
relate: https://github.com/milvus-io/milvus/issues/43120

---------

Signed-off-by: aoiasd <zhicheng.yue@zilliz.com>
2025-08-04 09:47:38 +08:00
congqixia
4aff581007
enhance: Pass callback in search batch pks to void large result (#43695)
Related to #43660

---------

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-08-02 17:57:37 +08:00
Buqian Zheng
01baf582d5
fix: GroupChunkTranslator to correctly identify vector field (#43706)
issue: #43653

Signed-off-by: Buqian Zheng <zhengbuqian@gmail.com>
2025-08-02 00:49:37 +08:00
Bingyi Sun
b59bc5e2c0
fix: make json path index non exists offsets compatible with 2.5 (#43691)
issue: https://github.com/milvus-io/milvus/issues/43666

---------

Signed-off-by: sunby <sunbingyi1992@gmail.com>
2025-08-01 23:22:23 +08:00
Buqian Zheng
b0226ef47c
fix: added more comprehensive container limit detection (#43693)
issue: #41435

Signed-off-by: Buqian Zheng <zhengbuqian@gmail.com>
2025-08-01 20:37:37 +08:00
Xianhui Lin
0f0edff7f0
fix: increment offset for null data rows in JsonKeyStats (#43679)
fix: increment offset for null data rows in JsonKeyStatsInvertedIndex
issue: https://github.com/milvus-io/milvus/issues/43151

Signed-off-by: Xianhui.Lin <xianhui.lin@zilliz.com>
2025-08-01 15:53:37 +08:00
congqixia
5f2f4eb3d6
enhance: Ignore entry with same ts when DeleteRecord search pks (#43669)
Related to #43660

This patch reduces the unwanted offset&ts entries having same timestamp
of delete record. Under large amount of upsert, this false hit could
increase large amount of memory usage while applying delete.

The next step could be passing a callback to `search_pk_func_` to handle
hit entry streamingly.

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-08-01 10:15:36 +08:00
Ted Xu
e37cd19da2
enhance: enable storage v2 by default (#43652)
Signed-off-by: Ted Xu <ted.xu@zilliz.com>
2025-08-01 08:59:36 +08:00
zhagnlu
239f743a18
fix: add enable_mmap key to load config (#43672)
#43670

Signed-off-by: luzhang <luzhang@zilliz.com>
Co-authored-by: luzhang <luzhang@zilliz.com>
2025-07-31 21:35:37 +08:00
sparknack
4aabe23a45
enhance: update flat_hash_map.hpp to a modified version (#43506)
issue: #41435

Signed-off-by: Shawn Wang <shawn.wang@zilliz.com>
2025-07-31 20:09:36 +08:00
congqixia
f29964bd17
fix: Add padding for sorted index preventing 0 length mmap (#43663)
Related to #43655

This patch add a padding when writing mmap file for ScalarSortedIndex in
case of mmap falure due to 0 mmap length.

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-07-31 18:53:36 +08:00
zhagnlu
708e426bb3
enhance: using set element for string term type (#43049)
issue: #43048

Signed-off-by: luzhang <luzhang@zilliz.com>
Co-authored-by: luzhang <luzhang@zilliz.com>
2025-07-31 10:35:37 +08:00
zhagnlu
31801f5937
fix: fix pk in [..] skip next batch when using multi-chunk segment (#43618)
#43494

Signed-off-by: luzhang <luzhang@zilliz.com>
Co-authored-by: luzhang <luzhang@zilliz.com>
2025-07-31 10:15:37 +08:00
congqixia
089f02bcca
fix: [StorageV2] Align null bitmap offset for fixed-length datatype (#43654)
Related to #43626

Similar to previous pr #43321, null bitmap could be dislocated if the
bitset ptr does not count the offset of array

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-07-31 09:55:36 +08:00
congqixia
6a74a7de66
enhance: Make DeleteRecord search pks by batch and PinAll (#43640)
Related to #43592

When delete records are large, search pk one by one will result into
many `Pincells` call which creates lots of futures.

This patch make search pk execute in batch to reduce this cost.

Also add `GetAllChunks` API to utilize `PinAllCells` to reduce pins.

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-07-30 19:15:36 +08:00
sthuang
a2c7ed2780
fix: [StorageV2] sort field binlogs paths for packed reader and writer (#43585)
key changes:
* fix unstable storage v2 compaction unit test by guaranteeing the order
of paths during sync.
* bump milvus-storage version, include
https://github.com/milvus-io/milvus-storage/pull/222
https://github.com/milvus-io/milvus-storage/pull/223
https://github.com/milvus-io/milvus-storage/pull/224
https://github.com/milvus-io/milvus-storage/pull/225
https://github.com/milvus-io/milvus-storage/pull/226
* Also fix the below related oom issue.
related: https://github.com/milvus-io/milvus/issues/43310

Signed-off-by: shaoting-huang <shaoting.huang@zilliz.com>
2025-07-30 08:09:36 +08:00
congqixia
4fe55e3008
fix: [StorageV2] Use separate channel for get_cells (#43632)
Related to #43584

There might be concurrent calls on `translator.get_cells`. The channel
cannot be shared among these calls, otherwise the logic will break.

This patch create new channel for each `get_cells` invocation in case of
data race.

---------

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-07-29 20:59:38 +08:00
foxspy
d57890449f
enhance: update knowhere version (#43528)
issue: #42937

Signed-off-by: xianliang.li <xianliang.li@zilliz.com>
2025-07-29 17:21:36 +08:00
Buqian Zheng
052fb6c562
feat: add time based eviction to data managed by cachinglayer (#43490)
issue: https://github.com/milvus-io/milvus/issues/41435

also added disk capacity protection

---------

Signed-off-by: Buqian Zheng <zhengbuqian@gmail.com>
2025-07-29 16:17:35 +08:00
Bingyi Sun
a765cd1eaa
enhance: unlink mmap file when chunk and index are destructed (#43524)
issue: https://github.com/milvus-io/milvus/issues/41636

---------

Signed-off-by: sunby <sunbingyi1992@gmail.com>
2025-07-29 16:05:36 +08:00
congqixia
268f1cdace
fix: Hold field shared_ptr in case of being released (#43614)
Related to #43584

Directly accessing `fields_` in `get_raw_data` may have race if load vec
index happens concurrently during getting raw data.

This PR make `bulk_subscript` hold shared_ptr of field column prevent
field column being release during reading it.

---------

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-07-29 12:15:36 +08:00
aoiasd
c9412434c8
enhance: add char group tokenizer (#42793)
relate: https://github.com/milvus-io/milvus/issues/42792
Add char group tokenizer which support use costum char group or use some
build-in char group as delimiters.

Signed-off-by: aoiasd <zhicheng.yue@zilliz.com>
2025-07-29 11:11:35 +08:00
congqixia
f666d89919
fix: [StorageV2] Access future result to get exception if any (#43613)
Related to #43584

When `LoadWithStrategy` throw exception, the ex was wrapped in the
returned future. If the future is not handled, this exception would be
ignored.

This patch add `future.get()` to get exception if any.

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-07-28 22:33:35 +08:00
Spade A
864d1b93b1
enhance: enable stlsort with mmap support (#43359)
issue: https://github.com/milvus-io/milvus/issues/43358

---------

Signed-off-by: SpadeA <tangchenjie1210@gmail.com>
2025-07-28 15:32:55 +08:00
zhagnlu
9bf1cb02d5
fix: add array_contains_all int to float converter (#43593)
#43334

Signed-off-by: luzhang <luzhang@zilliz.com>
Co-authored-by: luzhang <luzhang@zilliz.com>
2025-07-28 14:14:55 +08:00
Buqian Zheng
b497d3d7a4
fix: call promise->setValue only after released the ListNode mtx (#43547)
issue: #43261

`promise->setValue(folly::Unit());` may run callbacks inline and some of
them may attempt to grab `mtx_`. So we should not call
`promise->setValue(folly::Unit());` while holding the lock.

---------

Signed-off-by: Buqian Zheng <zhengbuqian@gmail.com>
2025-07-26 18:34:55 +08:00
Bingyi Sun
742d72a6c2
fix: Fix wrong null offsets for json path index (#43390)
issue: https://github.com/milvus-io/milvus/issues/43315

---------

Signed-off-by: sunby <sunbingyi1992@gmail.com>
2025-07-26 17:26:54 +08:00
Bingyi Sun
a89e579485
fix: use tantivy version to make json index compatible with milvus 2.5 (#43563)
issue: https://github.com/milvus-io/milvus/issues/43562

Signed-off-by: sunby <sunbingyi1992@gmail.com>
2025-07-26 17:18:55 +08:00
congqixia
0b860b4aec
fix: Revert "enhance: DataCodec to release ownership of input_data after initialization (#43542)" (#43571) 2025-07-25 20:48:16 +08:00
congqixia
2a7b7a811a
fix: [StorageV2] Throw exception when read rg fails (#43561)
Related to #43261

Read error with catched in `LoadWithStrategy`. Caller could not detect
read failure when some error occurred. This patch make
`LoadWithStrategy` throw ex instead of swallowing it.

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-07-25 17:40:55 +08:00
Buqian Zheng
d23205b718
enhance: DataCodec to release ownership of input_data after initialization (#43542)
issue: https://github.com/milvus-io/milvus/issues/43088
issue: https://github.com/milvus-io/milvus/issues/43038

see also https://github.com/milvus-io/milvus/pull/43533.

Signed-off-by: Buqian Zheng <zhengbuqian@gmail.com>
2025-07-25 14:24:54 +08:00
sthuang
5cebc9f7f6
fix: [StorageV2] handle correct cid with multiple files and add storage v2 prefix logs (#43539)
related: #43372

---------

Signed-off-by: shaoting-huang <shaoting.huang@zilliz.com>
2025-07-25 11:22:54 +08:00